CN104123284B - The method and server of a kind of recommendation - Google Patents
The method and server of a kind of recommendation Download PDFInfo
- Publication number
- CN104123284B CN104123284B CN201310145097.0A CN201310145097A CN104123284B CN 104123284 B CN104123284 B CN 104123284B CN 201310145097 A CN201310145097 A CN 201310145097A CN 104123284 B CN104123284 B CN 104123284B
- Authority
- CN
- China
- Prior art keywords
- user
- recommendation list
- recommendation
- commending system
- selection result
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
- H04N21/4668—Learning process for intelligent management, e.g. learning user preferences for recommending movies for recommending content, e.g. movies
Abstract
The embodiment of the present invention discloses the method and server of a kind of recommendation, methods described passes through the combined strategy that pre-sets, the recommendation list sent to each commending system is combined, realize that multiple recommended models or system give user feedback recommendation results, according to the select probability of user feedback, update the recommendation list after the combination, realize and assess in real time, and receive the selection result of user's periodic feedback, the recommendation list after the combination is updated, and then embodies the current interest and historical interest hobby of user simultaneously.
Description
Technical field
The invention belongs to the method and server of data processing field, more particularly to a kind of recommendation.
Background technology
In today of information explosion, increasing business system introduces recommended technology, from the pattern that content is looked for forefathers
It is transformed into content and looks for people, meets users ' individualized requirement.The effect of single commending system recommendation results is limited, and this point is particularly
It is increasing to recommend to be embodied in contest, the final award-winner of contest often using multiple recommended technology/models or
It is integrated that appraisal result carries out fusion.The more analogies of conventional recommendation have scholar to think recommendation list shape into prediction scoring problem
Formula may more properly, and the conventional bad assessment of commending system effect, is typically also offline progress, real-time recruitment evaluation is not added
To utilize.What history hobby represented is the hobby of user all the time, is usually to change within a very long time
, it can be obtained by analyzing the historical behavior of user;And currently like representative is the current interim hobby of user, one
As and the mutability with time and external environment.
In common prior art one, typically by importing data offline, user feedback is operation system to commending system
The real-time data feedback of user is to commending system, for updating recommended models to improve the forecasting accuracy in future.Conventional use
Family feedback system, there is collection, click on, browse(Time), purchase, marking, comment etc. behavior.The shortcomings that prior art one is,
Technical comparing is single, is difficult to take into account in terms of the accuracy of recommendation and the real-time of calculating.
In common prior art two, combination technique is recommended to obtain preference pattern by off-line training, it is substantially or single
One commending system.It is typically all strong correlation to recommend combination and rear end recommended technology, is disposed together.Recommend combination technique:God
Through network, case-based reason- ing(Case-based reasoning, CBR), decision tree etc..The shortcomings that prior art two, is, pushes away
The technology of recommending is not easy to extend, and often increasing a kind of recommended technology, then built-up pattern needs off-line training, and can not Real-time Feedback again
The current interest of user.
The content of the invention
It is an object of the invention to provide a kind of method of recommendation and server, there are multiple recommended models or system in solution
When, how recommendation list to be updated using real-time Evaluated effect, while embody the current interest and historical interest of user
Hobby.
In a first aspect, a kind of method of recommendation, methods described include:
Receive the recommendation list that each commending system is sent;
According to the combined strategy pre-set, the recommendation list is combined, the recommendation list after combination is presented
To user so that user is selected according to the recommendation list after the combination;
The selection result of user feedback is received, the recommendation list after the combination is updated according to the selection result.
With reference to the group that in a first aspect, in the first possible implementation of first aspect, the basis is pre-set
Strategy is closed, the recommendation list is combined, including:
The proportion that the result that pre-defined each commending system is sent accounts in the result that all commending systems are sent;
According to the proportion, the recommendation list sent to each commending system is combined.
With reference to the possible implementation of the first of first aspect or first aspect, second in first aspect may
Implementation in, the selection result for receiving user feedback, the recommendation after the combination is updated according to the selection result
List, including:
The selection result of user feedback is received, select probability is calculated according to the selection result;
Renewal coefficient is pre-set, is calculated according to the proportion of the select probability, renewal coefficient and each commending system each
Proportion after individual commending system renewal;
According to the recommendation list after the more Combination nova of the proportion after the renewal.
With reference to second of possible implementation of first aspect, in the third possible implementation of first aspect
In, the selection result for receiving user feedback, select probability is calculated according to the selection result, including:
The ratio of all results of recommendation list that the selection result is accounted for after the combination is calculated, the ratio is selection
Probability;Or
The weight of user's selection is pre-set, the weight selected according to the user obtains user and selects each commending system
The ratio of the weight for all result of recommendation list that the weight of the result of recommendation is accounted for after combination, the ratio is select probability.
Second with reference to first aspect either the first possible implementation or first aspect of first aspect can
The implementation of energy or the third possible implementation of first aspect, in the 4th kind of possible realization side of first aspect
In formula, methods described also includes:
The selection result of user's periodic feedback is received, according to the selection result of the periodic feedback, updates each recommendation system
The recommendation list of system.
Second aspect, a kind of server, the server include:
Receiving unit, the recommendation list sent for receiving each commending system;
Assembled unit, for according to the combined strategy pre-set, the recommendation list being combined, after combination
Recommendation list is presented to user so that user is selected according to the recommendation list after the combination;
Updating block, for receiving the selection result of user feedback, after updating the combination according to the selection result
Recommendation list.
With reference to second aspect, in the first possible implementation of second aspect, the assembled unit is specifically used for:
The proportion that the result that pre-defined each commending system is sent accounts in the result that all commending systems are sent;
According to the proportion, the recommendation list sent to each commending system is combined.
With reference to the possible implementation of the first of second aspect or second aspect, second in second aspect may
Implementation in, the updating block is specifically used for:
The selection result of user feedback is received, select probability is calculated according to the selection result;
Renewal coefficient is pre-set, is calculated according to the proportion of the select probability, renewal coefficient and each commending system each
Proportion after individual commending system renewal;
According to the recommendation list after the more Combination nova of the proportion after the renewal.
With reference to second of possible implementation of second aspect, in the third possible implementation of second aspect
In, the selection result that step receives user feedback is performed in the updating block, select probability is calculated according to the selection result,
Including:
The ratio of all results of recommendation list that the selection result is accounted for after the combination is calculated, the ratio is selection
Probability;Or
The weight of user's selection is pre-set, the weight selected according to the user obtains user and selects each commending system
The ratio of the weight for all result of recommendation list that the weight of the result of recommendation is accounted for after combination, the ratio is select probability.
Second with reference to second aspect either the first possible implementation or second aspect of second aspect can
The implementation of energy or the third possible implementation of second aspect, in the 4th kind of possible realization side of second aspect
In formula, the server also includes periodic feedback unit, is used for:
The selection result of user's periodic feedback is received, according to the selection result of the periodic feedback, updates each recommendation system
The recommendation list of system.
Compared with prior art, the present invention is by the combined strategy pre-set, the recommendation sent to each commending system
List is combined, and realizes that multiple recommended models or system give user feedback recommendation results, general according to the selection of user feedback
Rate, the recommendation list after the combination is updated, realize and assess in real time, and receive the selection result of user's periodic feedback, update institute
The recommendation list after combination is stated, because assessing the current interest that can embody user in real time, periodic feedback can embody user's
Historical interest, therefore the present invention can embody the current interest and historical interest hobby of user simultaneously.
Brief description of the drawings
Technical scheme in order to illustrate the embodiments of the present invention more clearly, it will use below required in embodiment
Accompanying drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the present invention, for ability
For the those of ordinary skill of domain, without having to pay creative labor, it can also be obtained according to these accompanying drawings other
Accompanying drawing.
Fig. 1 is a kind of application scenario diagram of the method for recommendation provided in an embodiment of the present invention;
Fig. 2 is a kind of method flow diagram of the method for recommendation provided in an embodiment of the present invention;
Fig. 3 is a kind of method schematic diagram of the method for recommendation provided in an embodiment of the present invention;
Fig. 4 is a kind of method schematic diagram of the method for recommendation provided in an embodiment of the present invention;
Fig. 5 is a kind of structure drawing of device of server provided in an embodiment of the present invention;
Fig. 6 is a kind of structure drawing of device of server provided in an embodiment of the present invention.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
All any modification, equivalent and improvement made within refreshing and principle etc., should be included in the scope of the protection.
With reference to figure 1, Fig. 1 is a kind of application scenario diagram of the method for recommendation provided in an embodiment of the present invention.As shown in figure 1,
The recommendation list that user 101 provides according to server 102, article interested etc. is selected from recommendation list, meanwhile, user
101 by the fructufy of selection when feed back to server 102, server 102 updates according to the selection result of user's Real-time Feedback is
Unite, the last hobby of user 101 can be reflected in the recommendation list that server 102 pushes to user 101 next time in time, and
And server 102 can periodically receive the selection result that user 101 feeds back, server 102 according to the selection result of periodic feedback more
New system so that server 102 is given the historical interest of energy simultaneous reactions user in the recommendation list that user 101 pushes and worked as every time
Preceding interest.
With reference to figure 2, Fig. 2 is a kind of method flow diagram of the method for recommendation provided in an embodiment of the present invention.As shown in Fig. 2
It the described method comprises the following steps:
Step 201, the recommendation list that each commending system is sent is received;
Specifically, as shown in figure 3, Fig. 3 is a kind of method schematic diagram of the method for recommendation provided in an embodiment of the present invention.Such as
Shown in Fig. 3, front end system is recommended to receive the recommendation list that commending system 1, commending system 2, commending system 3 are sent, the recommendation
The recommendation list that the commending system 1, commending system 2, commending system 3 are sent is carried out group by front end system according to combined strategy
Close, the recommendation list after combination is sent to operation system by the recommendation front end system so that after the operation system will combine
Recommendation list be presented to user.The select probability for recommending front end system to receive user's Real-time Feedback, it is described for updating
Recommendation list after combination, meanwhile, commending system 1, commending system 2, commending system 3 receive the selection knot of user's periodic feedback
Fruit, for updating the recommendation list of commending system 1, commending system 2, commending system 3, each commending system calculates according to database
Recommendation results it is different, such as commending system 1 is more it is recommended that articles for children, the electronic product that commending system 2 is more recommended
The either field such as books or clothes.
Step 202, according to the combined strategy pre-set, the recommendation list is combined, by the recommendation after combination
List is presented to user so that user is selected according to the recommendation list after the combination;
Alternatively, the combined strategy that the basis is pre-set, the recommendation list is combined, including:
The proportion that the result that pre-defined each commending system is sent accounts in the result that all commending systems are sent;
According to the proportion, the recommendation list sent to each commending system is combined.
Specifically, assuming that the recommendation list A that commending system 1 is sent to recommendation front end system is { a1, a2, a3 }, recommend system
The recommendation list B that system 2 is sent to recommendation front end system is { b1, b2, b3 }, and commending system 3, which is sent to, recommends pushing away for front end system
It is { c1, c2, c3 } to recommend list C.According to the combined strategy of the recommendation results pre-set, it is assumed that commending system 1, commending system 2,
When the proportion that the result that commending system 3 is recommended accounts for all commending system recommendation results is all 1/3, then the recommendation list after combining can
For { a1, b1, c2 }, it is assumed that the result that commending system 1, commending system 2, commending system 3 are recommended accounts for all commending systems and recommends knot
When the proportion of fruit is 3/5,1/5,1/5 respectively, then the recommendation list after combining can be { a1, a2, a3, b2, c3 }.
Meanwhile the proportion of combined strategy can be defined freely, it is assumed that during Children's Day, can be brought up again the ratio of commending system 1
Height, because the recommendation of commending system 1 is articles for children mostly.
Step 203, the selection result of user feedback is received, the recommendation after the combination is updated according to the selection result and is arranged
Table.
Alternatively, the selection result for receiving user feedback, according to pushing away after the selection result renewal combination
List is recommended, including:
The selection result of user feedback is received, select probability is calculated according to the selection result;
Renewal coefficient is pre-set, is calculated according to the proportion of the select probability, renewal coefficient and each commending system each
Proportion after individual commending system renewal;
According to the recommendation list after the more Combination nova of the proportion after the renewal.
Alternatively, the selection result for receiving user feedback, select probability is calculated according to the selection result, including:
The ratio of all results of recommendation list that the selection result is accounted for after the combination is calculated, the ratio is selection
Probability;Or
The weight of user's selection is pre-set, the weight selected according to the user obtains user and selects each commending system
The ratio of the weight for all result of recommendation list that the weight of the result of recommendation is accounted for after combination, the ratio is select probability.
Specifically, 3 commending systems in initialization Fig. 3, it is assumed that the proportion of each commending system is respectively P1(t)、P2(t)、
P3(t),
And
pi(t) accounting of i-th of commending system of t is represented
pi(t+1) accounting of i-th of commending system of subsequent time after expression t
η is renewal coefficient
λi(t) probability that the result that i-th of commending system of t is recommended is chosen by user is represented
Assuming that the recommendation list A that commending system 1 is sent is { i1, i2, i3 }, the recommendation list B that commending system 2 is sent is
{ i2, i3, i4, i5 }, commending system 3 do not send recommendation list, finally combination be presented to the recommendation list list of user for i1,
I2, i3, i4, i5 }, in order to avoid recommendation results assembled arrangement position on user select influence, we result is combined after with
Machine order of presentation.Assuming that user have selected { i2, i4 }, its real user selection is { i2 (A), i2 (B), i4 (B) }, so
Specifically, when user's operation behavior difference, such as:, behavior of the user to i2 is purchase, and the behavior to i4 is only
It is when browsing, then each select probability for recommending subsystem recommendation results to be chosen by user needs to consider each user's operation behavior
Weight, hence it is evident that buying behavior weight is greater than navigation patterns, it is assumed that buying behavior weight is 0.3 and navigation patterns weight is 0.2,
Then user's select probability is { i2 (A) * 0.3, i2 (B) * 0.3, i4 (B) * 0.2 }, so
To λA(T) for, 3/8 > 1/3, the recommendation effect of commending system 1 in the latter case is better than the first feelings
Condition, reason are exactly that the i2 that commending system 1 is recommended have purchased by user.
As another optional method, methods described also includes:
The selection result of user's periodic feedback is received, according to the selection result of the periodic feedback, renewal is described each to be pushed away
Recommend the recommendation list of system.
Specifically, commending system 1, commending system 2, commending system 3 periodically receive the selection result of user's periodic feedback, root
The recommendation list of oneself transmission is updated according to result.
By the combined strategy pre-set, the recommendation list sent to each commending system is combined the present invention, real
Existing multiple recommended models or system give user feedback recommendation results, according to the select probability of user feedback, update the combination
Recommendation list afterwards, realize and assess in real time, and receive the selection result of user's periodic feedback, update the row of the recommendation after the combination
Table, and then the current interest and historical interest hobby of user are embodied simultaneously.
With reference to figure 4, Fig. 4 is a kind of method schematic diagram of the method for recommendation provided in an embodiment of the present invention.Methods described bag
Include following steps:
Step 401, user first logs into, any information of system no user, then the recommendation front end system is in each recommendation
Recommendation results are chosen the recommendation list moderate of subsystem and are presented to user, or system recommends son according to combined strategy to be each
System assigns initial value;
Step 402, recommend front end system to obtain the recommendation list of each commending system and be combined displaying;
Step 403, user carries out selection to the result of recommendation and the operation such as browses or buy or collect;
Step 404, recommend front end system to capture the behavior of user in real time, assign different user behaviors different weight generations
The current hobby of table user, and the select probability for carrying out recommendation results calculates;
Step 405, front end system is recommended currently to be liked according to user by the defeated of certain each commending system of rule adjustment
The proportion gone out in final result list is recommended;
Step 406, user now can see the recommendation results before being different from;
Step 407, user log off terminates this session, each overall row for recommending subsystem to obtain this ession for telecommunication user
For;
Step 408, subsystem is respectively recommended to like adjustment recommendation list according to the history of user;
Step 409, user logs on system, jumps to step 401.
With reference to figure 5, Fig. 5 is a kind of structure drawing of device of server provided in an embodiment of the present invention.It is as shown in figure 4, described
Server is included with lower unit:
Receiving unit 501, the recommendation list sent for receiving each commending system;
Specifically, as shown in figure 3, Fig. 3 is a kind of method schematic diagram of the method for recommendation provided in an embodiment of the present invention.Such as
Shown in Fig. 3, front end system is recommended to receive the recommendation list that commending system 1, commending system 2, commending system 3 are sent, the recommendation
The recommendation list that the commending system 1, commending system 2, commending system 3 are sent is carried out group by front end system according to combined strategy
Close, the recommendation list after combination is sent to operation system by the recommendation front end system so that after the operation system will combine
Recommendation list be presented to user.The select probability for recommending front end system to receive user's Real-time Feedback, it is described for updating
Recommendation list after combination, meanwhile, commending system 1, commending system 2, commending system 3 receive the selection knot of user's periodic feedback
Fruit, for updating the recommendation list of commending system 1, commending system 2, commending system 3.
Assembled unit 502, for according to the combined strategy pre-set, the recommendation list being combined, will be combined
Recommendation list afterwards is presented to user so that user is selected according to the recommendation list after the combination;
Alternatively, the assembled unit 502, is specifically used for:
The proportion that the result that pre-defined each commending system is sent accounts in the result that all commending systems are sent;
According to the proportion, the recommendation list sent to each commending system is combined.
Specifically, assuming that the recommendation list A that commending system 1 is sent to recommendation front end system is { a1, a2, a3 }, recommend system
The recommendation list B that system 2 is sent to recommendation front end system is { b1, b2, b3 }, and commending system 3, which is sent to, recommends pushing away for front end system
It is { c1, c2, c3 } to recommend list C.According to the Generalization bounds combined in advance, it is assumed that commending system 1, commending system 2, commending system 3
When the proportion that the result of recommendation accounts for all commending system recommendation results is all 1/3, then combine after recommendation list can be a1, b1,
C2 }, it is assumed that the result that commending system 1, commending system 2, commending system 3 are recommended accounts for the proportion point of all commending system recommendation results
When not being 3/5,1/5,1/5, then the recommendation list after combining can be { a1, a2, a3, b2, c3 }.
Meanwhile the proportion of combined strategy can be defined freely, it is assumed that during Children's Day, can be brought up again the ratio of commending system 1
Height, because the recommendation of commending system 1 is articles for children mostly.
Updating block 503, for receiving the selection result of user feedback, after updating the combination according to the selection result
Recommendation list.
Alternatively, the updating block 503 is specifically used for:
The selection result of user feedback is received, select probability is calculated according to the selection result;
Renewal coefficient is pre-set, is calculated according to the proportion of the select probability, renewal coefficient and each commending system each
Proportion after individual commending system renewal;
According to the recommendation list after the more Combination nova of the proportion after the renewal.
The selection result that step receives user feedback is performed in the updating block, is calculated and selected according to the selection result
Probability, including:
The ratio of all results of recommendation list that the selection result is accounted for after the combination is calculated, the ratio is selection
Probability;Or
The weight of user's selection is pre-set, the weight selected according to the user obtains user and selects each commending system
The ratio of the weight for all result of recommendation list that the weight of the result of recommendation is accounted for after combination, the ratio is select probability.
Specifically, 3 commending systems in initialization Fig. 3, it is assumed that the proportion of each commending system is respectively P1(t)、P2(t)、
P3(t),
And
pi(t) accounting of i-th of commending system of t is represented
pi(t+1) accounting of i-th of commending system of subsequent time after expression t
η is renewal coefficient
λi(t) probability that the result that i-th of commending system of t is recommended is chosen by user is represented
Assuming that the recommendation list A that commending system 1 is sent is { i1, i2, i3 }, the recommendation list B that commending system 2 is sent is
{ i2, i3, i4, i5 }, commending system 3 do not send recommendation list, finally combination be presented to the recommendation list list of user for i1,
I2, i3, i4, i5 }, in order to avoid recommendation results assembled arrangement position on user select influence, we result is combined after with
Machine order of presentation.Assuming that user have selected { i2, i4 }, its real user selection is { i2 (A), i2 (B), i4 (B) }, so
Specifically, when user's operation behavior difference, such as:Behavior of the user to i2 is purchase, and the behavior to i4 is only
It is when browsing., then it is each to recommend the select probability chosen by user of subsystem recommendation results to need to consider each user's operation behavior
Weight, hence it is evident that buying behavior weight is greater than navigation patterns, it is assumed that buying behavior weight is 0.3 and navigation patterns weight is
0.2, then user's select probability is { i2 (A) * 0.3, i2 (B) * 0.3, i4 (B) * 0.2 }, so
To λA(t) for, 3/8>1/3, the recommendation effect of commending system 1 in the latter case is better than the first feelings
Condition, reason are exactly that the i2 that commending system 1 is recommended have purchased by user.
As a kind of optional embodiment, the server also includes periodic feedback unit 504, is used for:
The selection result of user's periodic feedback is received, according to the selection result of the periodic feedback, renewal is described each to be pushed away
Recommend the recommendation list of system.
Specifically, commending system 1, commending system 2, commending system 3 periodically receive the selection result of user's periodic feedback, root
The recommendation list of oneself transmission is updated according to result.
By the combined strategy pre-set, the recommendation list sent to each commending system is combined the present invention, real
Existing multiple recommended models or system give user feedback recommendation results, according to the select probability of user feedback, update the combination
Recommendation list afterwards, realize and assess in real time, and receive the selection result of user's periodic feedback, update the row of the recommendation after the combination
Table, and then the current interest and historical interest hobby of user are embodied simultaneously.
With reference to figure 6, Fig. 6 is a kind of structure drawing of device of server provided in an embodiment of the present invention.With reference to figure 6, Fig. 6 is this
A kind of server 600 that inventive embodiments provide, specific implementation of the specific embodiment of the invention not to the server limit
It is fixed.The server 600 includes:
Processor (processor) 601, communication interface (Communications Interface) 602, memory
(memory) 603, bus 604.
Processor 601, communication interface 602, memory 603 complete mutual communication by bus 604.
Communication interface 602, for being communicated with other equipment;
Processor 601, for configuration processor.
Specifically, program can include program code, and described program code includes computer-managed instruction.
Processor 601 is probably a central processor CPU, or specific integrated circuit ASIC(Application
Specific Integrated Circuit), or it is arranged to implement the integrated electricity of one or more of the embodiment of the present invention
Road.
Memory 603, for depositing program A.Memory 603 may include high-speed RAM memory, it is also possible to also including non-
Volatile memory(non-volatile memory), a for example, at least magnetic disk storage.Program can specifically include:
Receive the recommendation list that each commending system is sent;
According to the combined strategy pre-set, the recommendation list is combined, the recommendation list after combination is presented
To user so that user is selected according to the recommendation list after the combination;
The selection result of user feedback is received, the recommendation list after the combination is updated according to the selection result.
The combined strategy that the basis is pre-set, the recommendation list is combined, including:
The proportion that the result that pre-defined each commending system is sent accounts in the result that all commending systems are sent;
According to the proportion, the recommendation list sent to each commending system is combined.
The selection result for receiving user feedback, the recommendation list after the combination is updated according to the selection result,
Including:
The selection result of user feedback is received, select probability is calculated according to the selection result;
Renewal coefficient is pre-set, is calculated according to the proportion of the select probability, renewal coefficient and each commending system each
Proportion after individual commending system renewal;
According to the recommendation list after the more Combination nova of the proportion after the renewal.
The selection result for receiving user feedback, select probability is calculated according to the selection result, including:
The ratio of all results of recommendation list that the selection result is accounted for after the combination is calculated, the ratio is selection
Probability;Or
The weight of user's selection is pre-set, the weight selected according to the user obtains user and selects each commending system
The ratio of the weight for all result of recommendation list that the weight of the result of recommendation is accounted for after combination, the ratio is select probability.
Methods described also includes:
The selection result of user's periodic feedback is received, according to the selection result of the periodic feedback, renewal is described each to be pushed away
Recommend the recommendation list of system.
The preferred embodiment of the present invention is the foregoing is only, is not intended to limit the scope of the present invention..It is any
All any modification, equivalent and improvement made within the spirit and principles in the present invention etc., should be included in application claims
Within scope.
Claims (8)
- A kind of 1. method of recommendation, it is characterised in that methods described includes:Server receives the recommendation list that each commending system is sent;According to the combined strategy pre-set, the recommendation list is combined, the recommendation list after combination is presented to use Family so that user carries out selecting article interested according to the recommendation list after the combination, and the article includes articles for children Either books or clothes;The selection result of user feedback is received, the recommendation list after the combination is updated according to the selection result;The selection result for receiving user feedback, includes according to the recommendation list that the selection result is updated after the combination:The selection result of user feedback is received, select probability is calculated according to the selection result;Renewal coefficient is pre-set, each push away is calculated according to the proportion of the select probability, renewal coefficient and each commending system Recommend the proportion after system update;According to the recommendation list after the more Combination nova of the proportion after the renewal;The calculation formula of proportion after each commending system renewal is:Andpi(t) accounting of i-th of commending system of t is represented;pi(t+1) accounting of i-th of commending system of subsequent time after expression t;η is renewal coefficient;λi(t) probability that the result that i-th of commending system of t is recommended is chosen by user is represented.
- 2. according to the method for claim 1, it is characterised in that the combined strategy that the basis is pre-set, pushed away described List is recommended to be combined, including:The proportion that the result that pre-defined each commending system is sent accounts in the result that all commending systems are sent;According to the proportion, the recommendation list sent to each commending system is combined.
- 3. according to the method for claim 1, it is characterised in that the selection result for receiving user feedback, according to described Selection result calculates select probability, including:The ratio of all results of recommendation list that the selection result is accounted for after the combination is calculated, the ratio is general for selection Rate;OrThe weight of user's selection is pre-set, the weight selected according to the user obtains user and selects each commending system to recommend Result the ratio of the weight of all result of recommendation list that accounts for after combination of weight, the ratio is select probability.
- 4. according to the method described in claim 1-3 any one, it is characterised in that methods described also includes:The selection result of user's periodic feedback is received, according to the selection result of the periodic feedback, renewal each recommendation system The recommendation list of system.
- 5. a kind of server, it is characterised in that the server includes:Receiving unit, the recommendation list sent for receiving each commending system;Assembled unit, for according to the combined strategy pre-set, the recommendation list being combined, by the recommendation after combination List is presented to user so that user carries out selecting article interested, the article according to the recommendation list after the combination Including articles for children either books or clothes;Updating block, for receiving the selection result of user feedback, the recommendation after the combination is updated according to the selection result List;The updating block is specifically used for:The selection result of user feedback is received, select probability is calculated according to the selection result;Renewal coefficient is pre-set, each push away is calculated according to the proportion of the select probability, renewal coefficient and each commending system Recommend the proportion after system update;According to the recommendation list after the more Combination nova of the proportion after the renewal;The calculation formula of proportion after each commending system renewal is:Andpi(t) accounting of i-th of commending system of t is represented;pi(t+1) accounting of i-th of commending system of subsequent time after expression t;η is renewal coefficient;λi(t) probability that the result that i-th of commending system of t is recommended is chosen by user is represented.
- 6. server according to claim 5, it is characterised in that the assembled unit is specifically used for:The proportion that the result that pre-defined each commending system is sent accounts in the result that all commending systems are sent;According to the proportion, the recommendation list sent to each commending system is combined.
- 7. server according to claim 5, it is characterised in that step is performed in the updating block and receives user feedback Selection result, according to the selection result calculate select probability, including:The ratio of all results of recommendation list that the selection result is accounted for after the combination is calculated, the ratio is general for selection Rate;OrThe weight of user's selection is pre-set, the weight selected according to the user obtains user and selects each commending system to recommend Result the ratio of the weight of all result of recommendation list that accounts for after combination of weight, the ratio is select probability.
- 8. according to the server described in claim 5-7 any one, it is characterised in that the server also includes periodic feedback Unit, it is used for:The selection result of user's periodic feedback is received, according to the selection result of the periodic feedback, renewal each recommendation system The recommendation list of system.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310145097.0A CN104123284B (en) | 2013-04-24 | 2013-04-24 | The method and server of a kind of recommendation |
PCT/CN2014/075183 WO2014173237A1 (en) | 2013-04-24 | 2014-04-11 | Recommendation method and server |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310145097.0A CN104123284B (en) | 2013-04-24 | 2013-04-24 | The method and server of a kind of recommendation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104123284A CN104123284A (en) | 2014-10-29 |
CN104123284B true CN104123284B (en) | 2018-01-23 |
Family
ID=51768697
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310145097.0A Active CN104123284B (en) | 2013-04-24 | 2013-04-24 | The method and server of a kind of recommendation |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN104123284B (en) |
WO (1) | WO2014173237A1 (en) |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9810442B2 (en) | 2013-03-15 | 2017-11-07 | Google Inc. | Controlling an HVAC system in association with a demand-response event with an intelligent network-connected thermostat |
CN106776660A (en) * | 2015-11-25 | 2017-05-31 | 阿里巴巴集团控股有限公司 | A kind of information recommendation method and device |
CN106056231B (en) * | 2016-03-29 | 2021-04-23 | 北京四海道达网络科技有限公司 | Automatic matching recommendation method and device based on multi-party position and time availability |
CN105913286B (en) * | 2016-05-16 | 2020-02-11 | 达而观信息科技(上海)有限公司 | Method for automatically fusing multiple personalized recommendation models |
CN107734395A (en) * | 2017-11-07 | 2018-02-23 | 山东浪潮商用系统有限公司 | A kind of program commending method, device, computer-readable recording medium and storage control |
CN108322501B (en) * | 2017-12-22 | 2021-04-27 | 深圳创新科软件技术有限公司 | Message pushing method and device |
CN108419134B (en) * | 2018-02-05 | 2020-02-18 | 华南理工大学 | Channel recommendation method based on fusion of individual history and group current behaviors |
CN108763502B (en) * | 2018-05-30 | 2022-03-25 | 腾讯科技(深圳)有限公司 | Information recommendation method and system |
CN109063104B (en) * | 2018-07-27 | 2020-11-10 | 百度在线网络技术(北京)有限公司 | Recommendation information refreshing method and device, storage medium and terminal equipment |
CN109299368B (en) * | 2018-09-29 | 2020-11-24 | 北京思路创新科技有限公司 | Method and system for intelligent and personalized recommendation of environmental information resources AI |
CN109544276B (en) * | 2018-10-30 | 2022-07-15 | 珠海格力电器股份有限公司 | Product model selection method and processor |
CN111061956B (en) * | 2019-12-24 | 2022-08-16 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
CN113159810A (en) * | 2020-01-22 | 2021-07-23 | 百度在线网络技术(北京)有限公司 | Policy evaluation method, device, equipment and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101888515A (en) * | 2010-06-30 | 2010-11-17 | 中山大学 | Method and system for playing subscribed digital television programs |
CN102685565A (en) * | 2012-05-18 | 2012-09-19 | 合一网络技术(北京)有限公司 | Click feedback type individual recommendation system |
CN202711318U (en) * | 2012-04-09 | 2013-01-30 | 唐慈津 | Integrated network goods comparison system |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8566256B2 (en) * | 2008-04-01 | 2013-10-22 | Certona Corporation | Universal system and method for representing and predicting human behavior |
-
2013
- 2013-04-24 CN CN201310145097.0A patent/CN104123284B/en active Active
-
2014
- 2014-04-11 WO PCT/CN2014/075183 patent/WO2014173237A1/en active Application Filing
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101888515A (en) * | 2010-06-30 | 2010-11-17 | 中山大学 | Method and system for playing subscribed digital television programs |
CN202711318U (en) * | 2012-04-09 | 2013-01-30 | 唐慈津 | Integrated network goods comparison system |
CN102685565A (en) * | 2012-05-18 | 2012-09-19 | 合一网络技术(北京)有限公司 | Click feedback type individual recommendation system |
Also Published As
Publication number | Publication date |
---|---|
WO2014173237A1 (en) | 2014-10-30 |
CN104123284A (en) | 2014-10-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104123284B (en) | The method and server of a kind of recommendation | |
US11070643B2 (en) | Discovering signature of electronic social networks | |
CN104281961B (en) | For the advertisement in on-line system and the quality score system of content | |
CN104239385B (en) | For inferring the method and system of the relation between theme | |
Wang et al. | Opportunity model for e-commerce recommendation: right product; right time | |
CN103377250B (en) | Top k based on neighborhood recommend method | |
CA2843056C (en) | User-initiated boosting of social networking objects | |
CN105960656A (en) | Methods and systems for creating data-driven attribution model for assigning attribution credit to multiple events | |
CN103678518B (en) | Method and device for adjusting recommendation lists | |
Xiang et al. | Time-dependent models in collaborative filtering based recommender system | |
CN103049637B (en) | Strengthen the system and method for the content quality and user's participation of social platform | |
CN103295148B (en) | The method and apparatus generating and realizing data model | |
CN104281962B (en) | For the advertisement in on-line system and the united market of content | |
CN110796477A (en) | Advertisement display method and device, electronic equipment and readable storage medium | |
US20140279232A1 (en) | System and method for an affinity capture, user feedback and affinity analysis | |
CN107562875A (en) | A kind of update method of model, apparatus and system | |
CN108510303A (en) | Advertisement placement method and device | |
CN103744916B (en) | A kind of method and apparatus for sharing temperature information for being used to determine target video | |
CN106534902A (en) | Behavior analysis method and system | |
CN108632311A (en) | Information-pushing method and device | |
CN108764949A (en) | A kind of information-pushing method and equipment | |
CN107851261A (en) | For providing the method and system of relevant advertisements | |
CN106537437A (en) | Identifying media channels that have a high likelihood of multiple consumptions by one or more users | |
CN107077455A (en) | Flow mass is determined using the score traffic based on event | |
CN103475644B (en) | The method for pushing of a kind of network application and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |